This paper provides a novel, conceptually driven stance on the state of the contemporary analytic challenges faced in the treatment of dialogue as a form of data across on- and offline sites of learning. In prior research, preliminary steps have been taken to detect occurrences of such dialogue using automated analysis techniques. Such advances have the potential to foster effective dialogue using learning analytic techniques that scaffold, give feedback on, and provide pedagogic contexts promoting such dialogue. However, the translation of much prior learning science research to online contexts is complex, requiring the operationalization of constructs theorized in different contexts (often face-to-face), and based on different datasets and structures (often spoken dialogue). In this paper, we explore what could constitute the effective analysis of productive online dialogues, arguing that it requires consideration of three key facets of the dialogue: features indicative of productive dialogue; the unit of segmentation; and the interplay of features and segmentation with the temporal underpinning of learning contexts. The paper thus foregrounds key considerations regarding the analysis of dialogue data in emerging learning analytics environments, both for learning-science and for computationally oriented researchers.